Environ. Sci. Technol. 2006, 40, 7305-7311
Quantitative Determination of 1,4-Dioxane and Tetrahydrofuran in Groundwater by Solid Phase Extraction GC/MS/MS† CARL ISAACSON,‡ THOMAS K. G. MOHR,§ AND J E N N I F E R A . F I E L D * ,‡,| Department of Chemistry, Oregon State University, Corvallis, Oregon 97331, Santa Clara Valley Water District, San Jose, California 95118, and Department of Environmental and Molecular Toxicology, Oregon State University, Corvallis, Oregon 97331
Groundwater contamination by cyclic ethers, 1,4-dioxane (dioxane), a probable human carcinogen, and tetrahydrofuran (THF), a co-contaminant at many chlorinated solvent release sites, are a growing concern. Cyclic ethers are readily transported in groundwater, yet little is known about their fate in environmental systems. High water solubility coupled with low Henry’s law constants and octanol-water partition coefficients make their removal from groundwater problematic for both remedial and analytical purposes. A solid-phase extraction (SPE) method based on activated carbon disks was developed for the quantitative determination of dioxane and THF. The method requires 80 mL samples and a total of 1.2 mL of solvent (acetone). The number of steps is minimized due to the “in-vial” elution of the disks. Average recoveries for dioxane and THF were 98% and 95%, respectively, with precision, as indicated by the relative standard deviation of 1100 m2/g. Prior to extraction each 80 mL sample was spiked with 100 µL of 50 µg/mL dioxane-d8 and THF-d8 surrogates in blank domestic tap water (5 µg each). The disk was not pre-wetted with water or solvent. Aqueous samples were passed through the 25 mm disk under vacuum (67 kPa) and the disks were then dried by pulling laboratory air through the disk for 1 h. The disks were then removed from the holder and placed directly into 7306
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 23, 2006
2 mL autosampler vials followed by 1.2 mL of acetone and 9.4 µg of the butyl acetate instrumental standard. The vials were then capped and analyzed after 30 min and within 24 h. No handling problems were encountered during in-vial elution, such as activated carbon particles separating from the membranes. In addition to activated carbon, 2 poly(styrenedivinylbenzene) copolymers (SDB-RPS and SDB-XC), and a C18 bonded-phase silica (all from 3M Corporation, Minneapolis, MN) were selected for preliminary extraction experiments. The two SDB copolymers have a pore size of 80 Å, a particle size of 16 µm, and a surface area of 450 m2/g. The C18 disks have a pore size of 60 Å and a 20 µm particle size. The SDBRPS, SDB-XC, and C18 disks were prepared per manufacturer specifications. Breakthrough Experiments. Deionized water solutions of 1000 µg/L dioxane of 50, 80, 100, and 150 mL were extracted with each kind of disk. After extraction of samples, the activated carbon disks were dried under vacuum for 1 h, and all other disks were dried for 10 min. All disks were eluted directly in an autosampler vial with dichloromethane. These initial experiments used dichloromethane as the elution solvent; subsequent trials used acetone, a non-chlorinated solvent, which provided equal recovery. In order to conduct a mass balance on the disk extraction process, the water that passed through each disk was collected and analyzed via liquid-liquid extraction to determine the mass of dioxane that passed through the disk. Performance and validation of the liquid-liquid extraction is detailed in Supporting Information (SI). Spike and Recovery. The accuracy and precision of the activated carbon method were determined from experiments in which dioxane and THF were spiked into deionized water and into groundwater samples that were previously determined to contain no dioxane or THF (spike and recovery). The spike and recovery experiments were performed by spiking 50 µg of dioxane, 4 µg of THF, 5 µg dioxane-d8, and THF-d8 into each of five 80 mL aliquots of deionized water and blank site groundwater. The deionized water extracts were not analyzed for THF because the deionized water contained THF. Spike-addition experiments with non-blank groundwater samples were performed with samples composited from two wells. The spiked masses of dioxane and THF were selected to approximately double the baseline concentrations of dioxane and THF. The samples were extracted and analyzed by GC-MS/MS as described below. Spike-addition experiments were performed by spiking 625 µg/L dioxane, 50 µg/L THF, and 63 µg/L each of the dioxane-d8 and THF-d8 internal standards into each of five replicate 80 mL portions of groundwater that contained a baseline concentration of 600 µg/L dioxane and 50 µg/L THF. Blanks. Initial extracts of deionized water were observed to contain dioxane and THF. Analysis of the elution solvent, acetone, and the activated carbon disks eluted with acetone showed they were blank for dioxane and THF. However, when activated carbon disks were dried for 1 h under vacuum and then eluted in-vial, dioxane but not THF was detected. The dioxane was determined to come from hood air used in drying the disk. For this reason, all subsequent extractions were performed in a laboratory in which no dioxane or THF standard solutions were prepared or handled. THF was detected in preliminary experiments in which deionized “house” water was used to condition the activated carbon disk. The in-house deionized water was determined to contain approximately 2 µg/L THF. Wang et al. observed up to 13 000 µg/L THF in laboratory deionized water and they concluded that the glue used with PVC pipes was the source of the THF (35). The source of the THF contamination was determined to be the deionized water; the contamination
was eliminated by discontinuing conditioning the disk with deionized water. Eliminating this conditioning step had no adverse affects on the extraction process. To ensure that the extraction method did not lead to false positives, blank method extractions were performed along with each set of 10 field samples using deionized water for dioxane and blank water obtained from a domestic water supply for THF. Detection and Quantitation Limits. To determine the detection and quantitation limits of the method, single 80 mL samples of blank groundwater were spiked to give final dioxane concentrations ranging from 0.063 to 0.63 µg/L and THF ranging from 0.75 to 9.4 µg/L, which were then extracted as described above. The analyte concentration that gave a signal-to-noise ratio (S/N) of 3 was determined to be the detection limit and the concentration that gave a S/N of 10 was determined to be the quantitation limit. Gas Chromatography-Mass Spectrometry. Gas chromatography tandem mass spectrometry was used to identify and quantify dioxane, THF, dioxane-d8, and THF-d8. A Hewlett-Packard (Wilmington, DE) 5890 Series II gas chromatograph was used for all separations. The GC was equipped with a 30 m × 0.32 mm × 4 µm Supleco SPB-1 column (Bellefonte, PA). Splitless 1 µL injections were made at an inlet temperature of 150 °C. The initial oven temperature was 40 °C for 1 min then increased at 10 °C/min to 120 °C. The inlet pressure was pulsed to 170 kPa for 2 min then ramped down to 77 kPa in 10 s. The GC was coupled to a Finnigan Mat TSQ 700 mass spectrometer operated in positive chemical ionization (PCI) mode while using selected reaction monitoring. The transfer line, manifold, and collision cell temperatures were set at 265, 70, and 150 °C, respectively. Argon was the collision gas with collision offset energy of 15 eV, methane was the reagent gas, and the solvent delay was 4.25 min. Because of the limited fragmentation produced in PCI (only one transition observed), confirmation of the dioxane and THF was performed by analyzing samples by electron impact (EI) GC/MS. A Hewlett-Packard (Wilmington, DE) 5890 Series II gas chromatograph was coupled to a HewlettPackard 5972 Series mass selective detector. Splitless 1 µL injections were made at an inlet temperature of 150 °C. The initial oven temperature was 40 °C for 1 min then increased at 10 °C/min to 120 °C. The mass spectrometer was operated in electron impact mode (70 eV) and scanned 20-200 amu. Quantification. The recoveries of dioxane and THF were determined by peak area ratio to internal standards dioxaned8 and THF-d8, respectively. The six point calibration curves spanned concentrations of 0.31-3100 µg/L for dioxane and 3.1-300 µg/L for THF; all standards were prepared in acetone. Each calibration standard also contained 5 µg of dioxane-d8 and THF-d8 as internal standards and 9.4 µg of butyl acetate as the instrumental standard, respectively. In addition, the absolute recoveries of dioxane and THF were determined by ratioing their peak areas to that of the butyl acetate instrumental standard. An inverse squared weighting scheme was used to determine the linear regression for the calibration curve (Excel, Microsoft Corporation, Seattle, WA). All points on the calibration curve were within 20% (36) of the expected concentration and were retained for linear regression. Calibration curves were determined at the beginning of each sample set and blanks and check standards were run for each sample set. Overall, blanks and check standards comprised a minimum of 30% of the total samples analyzed during any given analysis sequence.
Results and Discussion Chromatography and Mass Spectrometry. The most abundant ions formed under positive chemical ionization were the [M + H]+ ions (26) and their collision-induced dissociation
FIGURE 1. Example of GC/MS/MS chromatograms of THF (13 µg/L), THF-d8 (63 µg/L), dioxane (300 µg/L), dioxane-d8 (63 µg/L), and instrumental standard butyl acetate in a groundwater sample. fragments were m/z ) 45 [C2H4O + H]+ for dioxane, m/z ) 49 [C2D4O + H]+ for the dioxane-d8, m/z ) 55 [C4H7]+ for THF, and m/z ) 62 [C4D7]+ for the THF-d8. The respective transitions were subsequently used for quantitation (Table S2). Chromatograms for samples containing dioxane, THF, dioxane-d8, and THF-d8 indicate that the deuterated internal standards elute earlier than do their nondeuterated analogs (Figure 1). PCI-GC/MS was selected over EI-GC/MS for quantification because PCI gave greater sensitivity than EI as determined from signal-to-noise determinations for replicate analyses of a single standard. Increased sensitivity under PCI conditions is consistent with the proton-accepting nature of these cyclic ethers. However, because only one mass transition was observed, samples were analyzed by EI-GC/ MS for confirmation. Dioxane under EI conditions produced ions including m/z 88 [M]+‚, m/z 58 [(CH2)3O]+‚, and m/z 28 [(C2H4)]+‚. The dioxane standard gave ion ratios ((95% confidence interval for n ) 3 replicate injections) of 0.45 ( 0.02 for m/z 88:m/z 28 and 0.28 ( 0.03 for m/z 58:m/z 28. Under EI conditions, THF gave ions including m/z 72 [M]+‚, m/z 71 [M+ - H]+, and m/z 42 [(CH2)3O]+‚. The THF standard gave ion ratios ((95% confidence interval for n ) 3 replicate injections) of 0.37 ( 0.06 for m/z 72:m/z 42 and 0.35 ( 0.03 for m/z 71:m/z 42. Analysis of dioxane (n ) 3) and THF (n ) 1) groundwater extracts by GC/MS under EI conditions gave ion ratios listed above that were within the 95% CI of the standards. In-Vial Elution. Acetone and dichloromethane were used as elution solvents and both provided greater than 90% recovery of dioxane from activated carbon disks. However, acetone was selected as the elution solvent of choice because it creates a less hazardous and more manageable laboratory waste. The time required to elute dioxane and THF from activated carbon disks was evaluated by performing repeated injections of a single sample. Recoveries greater than 90% were obtained after eluting for 30 minutes; this was the minimum time used to prepare all subsequent samples. Butyl acetate was determined to be an appropriate instrumental standard for the in-vial elution because no loss of butyl acetate onto the disks was observed during the elution process. SPE Optimization. Solid phases including SDB-RPs, SDBXC, and C18 gave maximum recoveries below 10% for 50150 mL samples of deionized water containing 1000 µg/L dioxane. The low recoveries were due to breakthrough because 82-110% of the added dioxane mass was recovered in water that had passed through the disk (Figure 2). Due to the unacceptably high degree of breakthrough, polymeric and silica-based phases were not used. In contrast, in a breakthrough study with 25 mm activated carbon disks, 92% recovery of dioxane (relative to the butyl acetate instrumental standard) was obtained for an 80 mL sample and recovery decreased to 52% for a 150 mL sample (Figure 2). All VOL. 40, NO. 23, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7307
TABLE 1. Accuracy ( Standard Error and Precision, as Indicated by Relative Standard Deviation, of the Activated Carbon Extraction Methoda deionized water spike recovery
blank groundwater spike recovery
groundwater spike additionc
recovery relative to instrumental standardb dioxane 93% ( 1 (3) 99% ( 1 (3) 86% ( 3 (7) dioxane-d8 94% ( 1 (3) 100% ( 1 (3) 85% ( 2 (6) THF NAe 82% ( 2 (5) 45% ( 2 (12) THF-d8 NA 83% ( 0 (1) 47% ( 2 (7) dioxane THF
recovery relative to internal standardd 100 ( 0 (1) 98% ( 1 (2) 100% ( 0 (1) NA 95% ( 2 (6) 100% ( 2 (4)
a Accuracy represented by average recoveries ( standard error and precision as indicated by the relative standard deviation of n ) 5 replicate analyses. b Recovery determined relative to butyl acetate instrumental standard. c Baseline concentrations of dioxane and THF were 600 and 50 µg/L, respectively. d Dioxane and THF recovery determined relative to their respective internal standards. e NA not analyzed because “house” deionized water was not blank for THF.
FIGURE 2. Breakthrough determined as a function of sample volume for (a) a 25 mm activated carbon disk and (b) a 25 mm SDB-XC disk. subsequent experiments were preformed with 25 mm activated carbon disks and an 80 mL sample volume. Once the suitability of the activated carbon method was determined for dioxane, the applicability of the method was determined for other solvent stabilizers including THF and 1,3-dioxolane. Spike and recovery experiments with THF in water from a domestic water supply gave greater than 80% recovery for THF and less than 10% for 1,3-dioxolane. Breakthrough of the 1,3-dioxolane was confirmed with 90% of the added mass recovered from the water that passed through the disk as determined by the in-vial liquid-liquid extraction method (described in SI). Therefore, 1,3-dioxolane cannot be determined by the activated carbon method. Activated Carbon Method Accuracy and Precision. Initial spike and recovery experiments into blank deionized water gave nearly equivalent absolute recoveries (93-94 ( 1%) for both dioxane and dioxane-d8 (Table 1). When dioxane recovery was computed relative to the dioxane-d8 internal standard, the relative recovery was 100 ( 0% (Table 1). Spike and recovery of THF into deionized water was not performed since the in-house deionized water contained THF at approximately 2 µg/L. Spike and recovery experiments with blank groundwater samples gave at least 99 ( 1% recovery for dioxane and 100 ( 2% recovery for dioxane-d8. Lower recoveries of THF and THF-d8 (82 and 83%) relative to dioxane (99%) were observed. When ratioed to their deuterated internal standards, both dioxane and THF gave recoveries that were at least 95% (Table 1). Spike-addition experiments, in which the background concentrations of dioxane and THF in a groundwater sample were doubled, gave average recoveries of dioxane and THF (relative to their deuterated internal standards) of 100 ( 0% and 100 ( 1%, respectively, with RSDs ranging from 1 to 4% (Table 1). Absolute recoveries (relative to the butyl acetate instrumental standard) of g85% were obtained for dioxane 7308
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 23, 2006
and dioxane-d8 and g45% for THF and THF-d8. The lower absolute recoveries of THF relative to dioxane may result from organic carbon loading of the activated carbon disk by other organic contaminants with higher Koc values present in the sample matrices. For example, the Koc for TCA is 2 orders of magnitude greater than the Koc for THF. Competitive adsorption of organic solutes on the activated carbon disk will favor those compounds with higher affinity to sorb to carbon, as well as those compounds present in higher concentrations (37). If dioxane or THF are present in heavily contaminated samples, competitive adsorption may result in reduced recoveries for dioxane and THF. However, the use of deuterated internal standards permits the detection of and correction for reduced recovery. The good agreement between the absolute recoveries of the deuterated and non-deuterated compounds and the high relative recoveries indicate the suitability of the deuterated internal standards for determining the concentrations of dioxane and THF in field samples. The low (
